Target Length: 3-5 minutes
Tone: Confident, technical, impressive
Objective: Win Track 1 by showcasing innovation, technical depth, and real-world utility
- Dramatic zoom into dashboard showing real-time threat detection
- Split screen: contract being exploited vs CodeGuard instantly pausing it
"In 2024, over $2 billion was lost to smart contract exploits. What if AI agents could detect and stop these attacks in real-time?"
"Meet CodeGuard - the world's first autonomous AI agent collective that protects your smart contracts 24/7."
"Built with Nullshot's Agent Framework and Model Context Protocol, this is decentralized AI at its finest."
- "CodeGuard: Autonomous Smart Contract Security"
- "Built for NullShot Hacks Season 0 - Track 1"
- Animated architecture diagram
- Show each component lighting up as mentioned
"CodeGuard isn't just one AI - it's a collective of six specialized agents working together:"
"The Orchestrator coordinates everything. The Monitor Agent watches every transaction in real-time. The Analyzer decompiles bytecode and runs static analysis. The Threat Intel Agent compares patterns against 500+ historical exploits using vector embeddings."
"When a critical threat is detected, the Response Agent automatically pauses the contract and alerts the team. And the Learning Agent? It makes the system smarter after every incident."
"All powered by 8 custom MCP tools - from bytecode analysis to mempool monitoring to semantic exploit search."
- "6 AI Agents"
- "8 MCP Tools"
- "Cloudflare Durable Objects"
- "Vector Search (500+ Exploits)"
- Navigate through the beautiful Next.js dashboard
- Show off the shadcn/ui components
- Real-time updates flowing in
"The dashboard is built with Next.js and shadcn/ui for a modern, polished experience."
"Here you can see all protected contracts with live risk scores. This circular gauge shows the overall security posture - currently looking good at 15."
"But watch this one - 72 out of 100. High risk. The AI agents are actively monitoring it."
"On the right, the live incident feed shows real-time detections via WebSockets. Every anomaly, every threat, streaming instantly."
"Down here, all five AI agents are healthy and operational - each with their own activity metrics."
- Highlight risk scores
- Point out WebSocket connection indicator
- Show agent status badges
- Terminal split screen
- Left: Triggering a re-entrancy attack
- Right: Dashboard reacting in real-time
- Show agent logs
"Now for the exciting part - let's trigger a re-entrancy attack on this vulnerable contract."
[Execute attack command]
"Watch what happens..."
"3 seconds - the Monitor Agent detects unusual gas usage patterns."
"5 seconds - the Analyzer Agent identifies the re-entrancy vulnerability."
"7 seconds - Threat Intel finds a 98% match to the infamous Cream Finance exploit from 2021."
"9 seconds - Response Agent executes emergency pause. Transaction confirmed."
"The contract is now frozen. Discord and Telegram alerts sent. Incident logged permanently onchain."
"Look at the dashboard - the incident appears in real-time with full details: contract address, threat type, risk score 95, and the AI agent that detected it."
- Timeline overlay: "3s → 5s → 7s → 9s → PAUSED"
- "Real-time Detection"
- "Autonomous Response"
- "Onchain Audit Trail"
- Show Basescan transaction
- Highlight smart contracts
- Show IPFS report
"Everything is transparent and verifiable on Base L2."
"Here's the pause transaction on Basescan - called by our AI agent's wallet."
"The GuardianRegistry contract tracks all protected contracts. The IncidentLog provides an immutable audit trail. And there's even an InsurancePool where users can stake to provide coverage."
"The full incident report? Stored on IPFS for permanent, decentralized access."
- "Base L2 (Low Cost, Fast)"
- "5 Smart Contracts"
- "Decentralized Storage (IPFS)"
- Show governance interface
- Highlight safety mechanisms
"Now, you might be thinking: 'That's a lot of power for an AI.' And you're right."
"That's why CodeGuard is governed by the community. The agent can pause contracts, but it can't access funds. There's a 1-hour veto window where the DAO can override any decision."
"To resume a contract? 24-hour timelock plus DAO approval required."
"This is decentralized AI done right - autonomous but accountable."
- "Community Governed"
- "1-Hour Veto Window"
- "24-Hour Timelock"
- "Funds Always Safe"
- Montage of features
- Show contract protection counter ticking up
- Fade to CodeGuard logo
"CodeGuard showcases everything Nullshot's framework enables: complex agent coordination, powerful MCP tools, and real Web3 integration."
"This isn't a concept - it's production-ready. Deploy it today and protect your protocols from the next billion-dollar exploit."
"CodeGuard: Because in Web3, security can't wait."
"Built for NullShot Hacks Season 0. Thank you."
- "CodeGuard"
- "Autonomous Smart Contract Security"
- "github.com/yourusername/codeguard"
- "Built with @NullShotFramework"
- "#NullShotHacks #Track1"
- Start with wide shots showing full dashboard
- Close-ups on key UI elements (risk gauges, agent status)
- Split screens for attack demonstration
- Smooth transitions between sections
- Futuristic, tech-focused background track
- Low volume so narration is clear
- Build tension during attack demo
- Triumphant during resolution
- Highlight boxes on important elements
- Animated arrows showing data flow
- Glow effects on real-time updates
- Code syntax highlighting in terminals
- Quick intro to hook viewers
- Steady technical explanation
- Exciting attack demo (climax)
- Confident closing
- Logos: Nullshot, Base, Cloudflare, OpenZeppelin
- Code snippets from key files
- Architecture diagrams
- Historical exploit news headlines
- Happy protocol team receiving alert
- Practice narration 3-5 times
- Prepare all demo environments
- Test all transitions
- Set up screen recording software (OBS/ScreenFlow)
- Configure microphone and test audio levels
- Prepare backup demo if live demo fails
- Record in 1920x1080 minimum
- Use high-quality microphone
- Record multiple takes of each section
- Keep raw footage organized
- Edit for pacing and clarity
- Add background music
- Add on-screen text and effects
- Color grade for consistency
- Export in MP4 (H.264)
- Upload to YouTube with unlisted link
- Test playback on multiple devices
- Include link in hackathon submission
- Add timestamps in video description
- Include relevant hashtags
- Share on Twitter with demo thread
- Show, don't just tell: Every feature claim should have a visual proof
- Build tension: The attack demo should feel dramatic
- Humanize the tech: Explain why this matters to real protocols
- Confidence: You built something amazing - own it!
- Call to action: End with clear next steps for judges
Every section directly addresses Track 1 criteria:
- MCP Integration: Repeatedly showcase 8 MCP tools
- Agent Coordination: Show 6 agents working together
- Decentralized AI: Emphasize governance and transparency
- Web3 Utility: Real-world security problem solved
- Innovation: Never been done before
- Technical Depth: Complex architecture clearly explained
- Production Quality: Polished UI and smooth demo
Remember: You have 5 minutes to convince judges that CodeGuard deserves to win Track 1. Every second counts. Make it count! 🚀